Dissolution kinetics of minerals in the presence of sorbing and

Cheng Fang Lin, and Mark M. Benjamin. Environ. Sci. Technol. , 1990, 24 (1), pp 126– ... Yun Wang and Alan T. Stone. Environmental Science & Technol...
0 downloads 0 Views 985KB Size
Environ. Sci. Technol. 1990, 2 4 , 126-134

(51) Low, M. J. D.; Hasegawa, M. J. Colloid Interface Sci. 1968, 26, 95-101. (52) Morrison, R. T.; Boyd, R. N. Organic Chemistry; Allyn and Bacon: Boston, MA, 1974. (53) Maes, A.; Leemput, L. V.; Cremers, A.; Uytterhoeven, J. J . Colloid Interface Sci. 1980, 77, 14-20. (54) Zerner, M. C. Ann. N.Y. Acad. Sci. 1981, 35, 367. (55) Pearlman, R. S.In Physical Chemical Properties of Drugs; Marcel Dekker: New York, 1972; pp 321-347.

(56) Pearlman, R. S.Quantum Chem. Prog. Exch. Bull. 1981, I , 15.

Received for review September 23, 1988. Revised manuscript received May 3,1989. Accepted August 8,1989. This research was supported by the Ecological Research Division, Office of Health and Environmental Research (OHER),US.Department of Energy (DOE),under Contract DE-AC06-76RLO 1830 as part of OHER’s Subsurface Science Program.

Dissolution Kinetics of Minerals in the Presence of Sorbing and Complexing Ligands Cheng-Fang Lin and Mark M. Benjamin* Department of Civil Engineering, FX-10, University of Washington, Seattle, Washington 98 195

This study investigated and modeled the important reactions controlling oxide dissolution in a system containing strongly complexing and strongly sorbing ligands. Polyphosphate and ferrihydrite were the model ligand and oxide studied, respectively. The dissolution of ferrihydrite in these systems is via a process called ligand-promoted dissolution, which is controlled by a surface reaction and is initiated by sorption of tripolyphosphates. Fe and tripolyphosphate leave the surface as one entity. If empty surface sites are available, Fe-tripolyphosphate complexes that are released to solution can quickly readsorb to the surface, forming a species different from the precursor to the dissolution reaction. The strong adsorption of complexes counteracts the iron dissolution reaction and leads to nonlinear net dissolution kinetics. An adsorption/ dissolution kinetic model is developed that successfully simulates the experimental observations here as well as previous work reported in the literature for systems where adsorption of free ligands and complexes is much weaker.

Introduction The surfaces of iron and aluminum oxides (and hydroxides) are capable of coordinating with (i.e., adsorbing) dissolved anions via ligand-exchange reactions (1-3). Several studies have shown that the solubility and the rate of dissolution of sparingly soluble oxides can be significantly increased by certain sorbing anions and organic substances (4-7). The increased rate of dissolution has been attributed to surface processes initiated by the sorbed species (8-10). Although the dissolution rate of these oxides may at times be limited by solid- or liquid-phase transport processes (11-16), Stumm and co-workers (16, 17) concluded that in most cases the rate of dissolution of slightly soluble oxides is controlled by the rate of surface chemical reactions. Some dissolution processes include reduction of the structural metal ion; in such cases dissolution involves the adsorption of reductants, precursor complex formation, electron transfer, release of oxidized adsorbate (anionic ligand or organic substance), and the release of reduced structural metal ion (10, 18). The reaction steps for nonreductive dissolution can be represented as follows (6): (1) adsorption of ligand Fe-O-Fe-1-OH

+ L = Fe-O-Fe-1-L + OH-

(2) formation of precursor complex

Fe-O-Fe-1-L 126

= Fe-O-1-FeL

Environ. Sci. Technol., Vol. 24, No. 1, 1990

(3) detachment of metal ion and generation of new site

Fe-O-1-FeL

+ H20

-

Fe-[-OH

+ FeL + OH

Once ligands adsorb to the surface, they may begin to polarize, weaken, and finally break the metal-oxygen bonds, causing the release of metal into solution. The overall process is referred to as “ligand-promoted dissolution” (6). In the absence of anionic ligands and under acidic conditions, surface hydroxyl groups are protonated. These protonated surface OH groups can also initiate the dissolution process in a way analogous to that caused by sorbed ligands. This is called “proton-promoted dissolution” (6). Stumm and co-workers (6, 7) combined the surface complexation model of adsorption with the concepts of ligand-promoted dissolution to develop an overall model of the dissolution process. Specifically, the ligand-promoted dissolution rate of an oxide is presumed to be directly related to the concentration of “precursor complexes” on the surface: RL = k [Me-O-1-MeL], where RL, k , and [Me-O-1-MeL] denote the ligand-promoted dissolution rate, a first-order rate constant, and the concentration of precursor complex, respectively. They suggested that in general the concentration of precursor complexes (which are not amenable to direct experimental analysis) will be proportional to the concentration of adsorbed ligands (which is easily measured). In such cases, RL = KLCsL,where KL and CSLdenote the ligand-promoted dissolution rate constant and the sorbed ligand concentration. If sorption of the ligand is fairly weak, CsLwill be small and a large number of surface sites will be vacant, even when the soluble ligand concentration is large. Under these conditions, CsLattains a pseudo steady state, and a constant dissolution rate can be established and maintained in a system for a fairly long period of time. Despite the undeniable importance of linear ligandpromoted dissolution kinetics (RL= KLCsL)in many systems, a number of studies have shown that under certain conditions the rate of dissolution decreases as dissolution proceeds and that the relationship between the dissolution rate and the adsorbed ligand concentration can be nonlinear ( 4 , Ei, 9,19). These observations have been attributed to a decrease in oxide surface area, heterogeneity of the surface, and the adsorption of released metal ions (18). The majority of the studies described above were conducted with weakly sorbing ligands. The controlling reactions might be quite different in systems with strongly sorbing ligands. The current study investigated the applicability of available models in systems of strongly

0013-936X/89/0924-0126$02.50/0

0 1989 American Chemical Society

I ill

12s

-

100

-

nii

-

I

PH 5

I

I

n ?i OW

n 0

1

3

4

5

I

3

7

4

6 Dissolved PjOlO (ri0-3M)

TOMI

moio (XIO.'M)

Figure 1. Dissolved Fe after 2-h contact of ferrihydrite (lo-' M as Fe) with tripolyphosphate. Dissolution does not occur until the tripolyphosphate concentration is M.

Figure 2. Concentrations of soluble Fe and tripolyphosphate after 7days contact with ferrihydrite at pH 5. 0.1 I

D D

0 10

sorbing ligands. As will be shown, the sorption of released metal-ligand complexes can be extremely important in such systems and, in fact, can control the overall kinetics of dissolution. In this paper, the experimental evidence supporting the importance of this reaction is presented, and the adsorption/dissolution kinetic model is expanded to include the reaction and explore the conditions under which it must be considered.

Experimental Section All chemicals used in this study were analytical grade. Tripolyphosphate was the strongly sorbing ligand investigated. Stock solutions of 0.1 and 0.01 M sodium tripolyphosphate (Na5P3010)were prepared and stored at 4 "C in the dark. Preparation of Ferrihydrite. The slightly soluble oxide chosen for study was ferrihydrite, which was prepared in batch for each experiment by adding ferric nitrate stock solution to a 0.1 M NaN03 solution at room temperature (22 f 2 "C) under a nitrogen atmosphere. The solution was rapidly titrated to pH 8.0 with NaOH, and the slurry was aged at pH 8.0 f 0.2 for 2 h before beginning the dissolution experiments. Dissolution Kinetics. Dissolution experiments were carried out at room temperature and at 1.5 "C. Ferrihydrite concentration was M Fe, and total tripolyphosphate concentrations were from 5 X to 2 X M. Both the suspension and stock tripolyphosphates were preadjusted to pH 5. After the ferrihydrite and tripolyphosphate were mixed, system pH was kept at 5.0 by using a Metrohm autotitrator. Normally, the dissolution reaction was allowed to proceed for 4 h. Sample aliquots were withdrawn at predetermined intervals and filtered immediately through 0.1-pm membrane filters; soluble Fe was operationally defined as the total Fe in the filtrate. Longer term (7 days) experiments were also conducted to investigate the stoichiometric relationship between soluble Fe species and tripolyphosphate species at equilibrium conditions. For these experiments, the system temperature was 1.5 "C and pH was 5.0. Fe and Phosphorus Analyses. Analyses of metals and phosphorus all followed recommended procedures (20). Fe concentrations in solution were determined by inductively coupled plasma (ICP) spectrometry. Both the standards and sample dilutions were made in 0.1 M NaN03 or 0.01 M NaN03. All solutions were adjusted to pH less than 3 prior to analysis. Polyphosphates were hydrolyzed to orthophosphate by boiling with acid and potassium per-

4

P

1

0

0.09 0.08

1

0.07

't

0.05

1

8

0.04

-I

4

0.03

-

0

0.02

-

0.01

-

-

d

0.06

P

4

2

0.00

1

I

-'

2 Y

I

M I

3 0

2

4

6

R

10

12

14

16

18

20

Equilibrium Soluble P3010 (~10-3M)

Figure 3. Adsorption isotherm of tripolyphosphate on ferrihydrite at pH 5. The maximum surface P3OlOconcentration is -0.1 mol of P3010/molof Fe and is attained at very low soluble P3O10 concentrations.

sulfate for 30 min in an autoclave, after which phosphorus was determined colorimetrically following the ascorbic acid method (Standard Methods 424F). A Perkin-Elmer Lambda 3 UV/vis spectrophotometer was used for these measurements.

Results Our previous study on the effect of polyphosphates on metal sorption behavior (21) revealed that the dissolved Fe concentration after 2 h is approximately proportional to the total tripolyphosphate added, as shown in Figure 1. However, dissolution did not occur in that system until the tripolyphosphate concentration was greater than M, corresponding to a ratio of -0.1 mol of total tripolyphosphate/mol of Fe. The importance of the ligand in this process is apparent from Figure 2, which shows the soluble Fe concentration as a function of soluble tripolyphosphate concentration at pH 5 after 7 days of contact. The plot is linear, passing through (0,O) with a slope of 0.98. Since no iron dissolves under similar conditions in the absence of tripolyphosphate, it is reasonable to conclude that essentially every iron ion in solution is complexed by tripolyphosphate, and vice versa. The adsorption isotherm for a tripolyphosphate/ferrihydrite system is shown in Figure 3. Adsorption of P3O10 is very strong, and almost all the P3OlOin the system is adsorbed until the surface is saturated (rpmax = 0.1 mol of P,Olo/mol of Fe), after which P3O10 can be detected in solution.

-

Environ. Sci. Technol.. Vol. 24. No. 1. 1990

127

in

2.5

Ill

I

i

ox

.'"4

IAl

0 0 I

0 U 0 0

I

0

00

02

04

06

08

10

I2

A d s o r t a I P 3 0 1 0 ( ~ 1 OM - ~I

Figure 4. Dependence of dissolution rate on the concentration of adsorbed tripolyphosphate in the system. Reaction time is 1.0 min. Note that the maximum surface P3OlO concentration is mol/L, corresponding to the maximum adsorption density (0.1 mol of P3Of0/molof Fe) times the Fe concentration in the system, 1 X lo-* M. The rate is clearly not linearly proportional to Cs,.

-

I?

so

25

0

IM

5

1

0

0

A

A

According to the ligand-promoted reaction model, dissolution of oxides is initiated by adsorption of anions through surface complexation reactions, and the rate of dissolution is approximately proportional to the adsorbed ligand concentration ( C s ~ ) By . contrast, Figure 1 and 3 indicate that in our system dissolution is negligible as the adsorbed ligand concentration increases from 0 to its saturation value. Thereafter, dissolution increases steadily as more polyphosphate is added to the system, even though CsLremains approximately constant. The average dissolution rates for the first minute of reaction are plotted as a function of the concentration of adsorbed P3Ol0in Figure 4; the nonlinearity of the relationship is obvious. Figure 5 shows the kinetics of ferrihydrite dissolution M tripolyphosphate in the presence of 5 X lo-" to 4 X at pH 5. In systems containing 5 X lo4, 8 X lo-", and M tripolyphosphate (Figure 5A), total soluble Fe concentration increases in the first 1,4, and 15 min, respectively, and decreases thereafter. [A similar pattern was observed at pH 7,8, and 10 for systems containing 8 X lo-" M P3Ol0 (21).] A t higher total tripolyphosphate concentrations (Figure 5B), dissolution kinetics appear to follow a nonlinear rate law, and no longer term decrease in soluble Fe concentration is observed. It is obvious from these two figures that the rates of ferrihydrite dissolution in the systems studied are not constant; rates of dissolution decrease as reaction proceeds, and in some cases the net dissolution rate becomes negative. Continually decreasing dissolution rates over time have been reported before (18,19) and have been attributed to reductions in surface area and to surface heterogeneity. Although these explanations might be appropriate for other systems, they cannot explain the results in the current study. Neither a decrease in surface area nor surface heterogeneity could cause Fe to return to the surface after it had dissolved, as shown in Figure 5A. The results imply that there is a reaction opposing the detachment step that can lead to a net reversal of the dissolution reaction. To our knowledge, no experimental evidence of such a reaction has been presented previously. For the time scale of these experiments, this reaction is significant and the decrease in dissolved Fe concentration is obvious only at the lower tripolyphosphate concentrations studied. At high tripolyphosphate concentrations, this "reverse dissolution" reaction may be overwhelmed 128

Environ. Sci. Technol., Vol. 24, No. 1, 1990

I50

I25

n

21

10

11

Iw

125

150

175

2W

Time 0 "

to 4 Figure 5. Dissolution kinetics at pH 5 in systems with 5 X X M tripolyphosphate. An increase followed by a decrease in dissolved Fe concentration is evident at total P3OlOless than or equal to 10-3 M.

by the detachment step, so that the net change in Fe concentration in those systems is positive at all times. There are two ways to account conceptually for the loss of dissolved Fe from solution: precipitation and adsorption. Since the dissolution of ferrihydrite is promoted by surface coordination reactions between tripolyphosphates and structural iron ions, iron is presumed to leave the surface together with tripolyphosphate as a complex. The fate and ultimate concentration of iron in solution depend on the stability constants and the sorption behavior of these complexes. The released Fe-tripolyphosphate complexes might dissociate, allowing free iron to reprecipitate or adsorb, or the complexes might adsorb and remove Fe from solution, if surface sites are available and if the adsorptive reaction is favorable. While either scenario (dissociation of the complexes followed by sorption or reprecipitation of the Fe; or adsorption of the complexes) is plausible conceptually, the former is not consistent with the experimental results, as shown below. Consider a set of reactions in series: A + B 11 1, ... I, P in which A and B are reactants, Ii are intermediates, and P is a product. For a system initially containing only A and B, the time profdes for the concentrations of all species are shown in Figure 6. While the concentrations of the species Ii may increase, peak, and then decrease, that of P can only increase toward its equilibrium value; it can never exceed that value, even transiently. This is true even if the reaction forming P is reversible, or if P can react to form A, B, or any intermediate directly. With respect to the system under consideration, the reactants A and B are surface structural iron atoms and ++

-

-

Table 1. Kinetic Expressions for the Proposed Dissolution Reactions of Ferrihydrite in the Presence of Tripolyphosphate (Figure 6 ) O

surface species d[Fe- -1- -L]/dt = kl[S1] [L] - k2[Fe--1- -L] - k3[Fe-l-L] + k4[Fe-O- -1- -FeL] d[Fe-0- -1- -FeLl/dt = k,[Fe- -1- -L] - k4[Fe-O- -1- -FeL] - k5[Fe-0- -1- -FeL] + k,[Sl][FeL] d[Fe-l-LFel/dt = kll[Sl][FeL] - klz[Fe-l-LFe] solution species d[FeLl/dt = k5[Fe-O--l--FeL] - k,[FeLI - k7[FeL] + k8[Fe][L]- kll[S1][FeL] + k,,[Fe-l-LFe] d[L]/dt = -kl[S1][Ll + k,[Fe--l--LI + k,[FeL] - k8[Fe][L] d[Fe]/dt = k,[FeL] - k8[Fel[Ll + ks([S1] + [S2]) - klo([S1] + [S2])[Fe] ligand-active sites d[Sl]/dt = -kl[S1][L] + k2[Fe--l--L]+ k5[Fe-O--l--FeL] - k,[FeL] - kll[FeL] + klz[Fe -I-LFe] H*-active sites d[S2]/dt = ks[S1] - klo[Fe][S1] [Sl] represents the concentration of ligand-active sites; [S2] represents sites that can undergo proton-promoted dissolution only.

reaction overwhelms the forward reaction. As a result, the Fe concentration in solution decreases. On the other hand, if the total tripolyphosphate concentration in the system exceeds the amount necessary to saturate the surface, the surface lacks capacity for all the dissolved Fe-tripolyphosphate complexes to adsorb. Both the forward and reverse reactions proceed; however, the forward reaction prevails, resulting in apparent nonlinear kinetics (Figure

5B). Time

Flgure 6. Time profiles for the concentrations of all species for a system initially containing only A. I,,,, A, and P, represent the equilibrium concentrations for the various species.

dissolved polyphosphate, and the intermediate species might be adsorbed polyphosphate, precursor complexes, etc. For the concentration of dissolved Fe complexes to increase and later decrease, the dissolved complex must be an intermediate that reacts to form a product different from any of the species that preceded it in the reaction sequence. A scenario in which the complexes dissociate and Fe returns to the surface via a sorption or precipitation reaction does not meet this requirement. In such a case the concentration of dissolved complex could increase, but could never decrease as it was observed to do. We conclude that the reaction sequence includes dissolved Fe-polyphosphate complexes as intermediate species that subsequently adsorb to form a new surface species, different from any of the precursor species. Since all the Fe in solution is apparently in complexed form and since other metal-tripolyphosphate complexes have been shown to adsorb strongly to ferrihydrite (21),this scenario seems entirely plausible. In the subsequent discussion, we suggest a possible structure for such a surface species. The kinetic results observed here can be interpreted by modifying the ligand-promoted dissolution model of Stumm et al. (22) to include the adsorption of Fe-tripolyphosphate complexes. The proposed sequence of events is that the detachment of metal from the surface is initiated as ligands adsorb. The soluble iron concentration at any time represents the net effect of a forward reaction (release of Fe complexes into solution) and a backward reaction (return of the complexes to the surface). Which reaction prevails depends on the relative concentration of surface sites and tripolyphosphate. The results presented in Figure 5 clearly indicate that adsorption of tripolyphosphate is fast and leads to rapid release of iron from the surface (the shortest sampling time is 1 min). If the total tripolyphosphate in the system is less than the amount required to saturate the surface, Fe-P,Olo complexes can readsorb, and after a few minutes, the backward

The essential aspect of the above scenario is that FeP,Olo complexes must adsorb to form a surface species different from those that led to Fe dissolution. We have also argued that Fe is initially released to solution as a complex by a ligand-promoted reaction. The possibility that Fe originally enters solution as uncomplexed, free Fe3+ ions released from bare surfaces sites, followed by very rapid coordination with dissolved tripolyphosphate and adsorption of complexes, cannot be absolutely ruled out. If this were the case, however, the overall dissolution process would be controlled by the proton-promoted dissolution reaction. On the basis of the proton-promoted dissolution rates of other iron oxides (7, 23), it is highly unlikely that such a reaction could dissolve 6% of M ferrihydrite in 2 h, as occurred in the system with 2 X 10” M tripolyphosphate. Thus, we conclude that release of F e P 3 0 1 0complexes from the surface, followed by their readsorption to form a new surface species, is the most likely sequence of events occurring in the systems studied. In the next section, the adsorption/dissolution kinetic model of Stumm and co-workers (6, 7) is extended to incorporate this readsorption reaction.

Model Development and Verification The proposed dominant reaction sequence for the dissolution of ferrihydrite in the presence of polyphosphate is (1) adsorption of tripolyphosphate, (2) formation of precursor complexes from sorbed tripolyphosphates, (3) release of precursor complexes, and (4) adsorption of complexes. After Fe-tripolyphosphate is released to solution, it will undergo dissociation and association reactions. In addition to the detachment of iron by release of precursor complexes, Fe may be released from vacant surface sites via proton-promoted dissolution and may reprecipitate. A schematic representation of the relevant reactions is presented in Figure 7, in which the forward and reverse reaction rate constants are noted. The kinetic expressions for the reactions in Figure 7 are grouped in Table I, where P3Olo is represented as L. As noted earlier, when FeL adsorbs it must form a surface species different from the dissolution precursor. Such a species might have the ligand bound directly to the surface, although other configurations are also possible. In this conceptual model, rapid adsorption of L leads to Environ. Sci. Technol.. Vol. 24, No. 1, 1990 129

kl

Q I

Fe-0- ----FeL

FeL

k7 + -ks

Fe

+

L

Fe-0-Fe.-----LFe

0

Observation Simulation

Flgure 7. Schematic representation of the reactions for the dissolution of ferrihydrite in the presence of tripolyphosphate. L represents tripolyphosphate. k , is the rate constant. The precursor complex (Fe0-I-FeL) is derhred from the adsorbed r i n d (Fe-0-Fe-I-L). Dissolved FeL can adsorb to form a surface complex (Fe-0-Fe-I-LFe) different from the one in the dissolution sequence.

0

50

100

150

200

Time (min)

Figure 8. Optimization results for system with lo-* M ([Fe],) ferrihydrite and 2 X M tripolyphosphate.

Table 11. Assumptions for Adsorption/Dissolution Kinetic Model Table 111. Optimized Kinetic Constants Used To Model Experimental Data

1. [SI] = o.lO[Fe]~ = 0.77[Fe]~ 2.

3. a surface reaction is the rate-determining step

12,

4. the precursor complex, Fe-0-1-FeL, is derived

k2 k, k, k, kR

from surface ligands

the formation of a weakly adsorbed surface FeL complex, which is then released to solution. Once in solution, FeL can return to the surface as a strongly sorbed complex. In particular, if the complex does adsorb with the polyphosphate attaching to the surface, it is reasonable to assume that adsorption of the complex and adsorption of free ligand are competing reactions. If the concentration of free L in solution is low and the adsorption density of L is less than rpmar, FeL can adsorb, decreasing the total soluble iron concentration; however, if free dissolved L is large, return of the complex to the surface could be impeded. Several assumptions were made before the kinetic rate constants of this model were optimized. The assumptions are listed in Table 11. The total concentration of adsorbing sites ([Sl])used in the model was O.1[FeIT. This number is obtained from the adsorption isotherm of tripolyphosphate. [Sl] + [S2] is the proton-exchange capacity of ferrihydrite, which is -0.87 mol/mol of Fe, based on the results of Davis and Leckie (24). The third assumption is the central assumption of surface reaction controlled dissolution. Zutic and Stumm (16) reported that for amorphous oxides and anions with small molecular size like F-, both diffusion and the surface reactions have to be considered in the rate-determining step. However, if a ligand has a molecular size larger than F-, as in the current case, the detachment step is more likely to be critical (16). Thus, it is justified to assume that a surface reaction is the rate-determining step. The final assumption is that the precursor complexes, which relate directly to the rate of dissolution, are derived from adsorbed uncomplexed ligands only. An adsorbed iron-tripolyphosphate complex, Fe-1-LFe, may lead to the formation of a binuclear precursor complex (Fe-0-Fe-1-LFe Fe-0-1FeLFe), but such a complex would probably be less capable of facilitating the detachment step. Computer codes based on the kinetic expressions in Table I were attached to NPSOL, a nonlinear optimization programming computer package (25). Kinetic results from the system with 2 X M tripolyphosphate and M ([FeIT) ferrihydrite at pH 5 were then used to optimize the rate constants in the model.

-

130

Environ. Sci. Technoi., Vol. 24, No. 1, 1990

1.10 X 1.25 X 4.93 X 2.84 X 8.95 X 6.45 X

lo3 M-' min-' lo4 min-' IO-' min-' lo2 min-' 10' min-' 10' M-' min-'

k, k, k9 k,, kll k,,

3.22 X 2.22 X 7.32 X 2.81 X 5.77 X 1.26 X

min-' lo4 M-' min-' min-' lo4 M-' min-' lo3 M-' min-' lo-*min-'

Optimization results are presented in Figure 8 and Table 111. Model simulations fit the experimental data very well. The optimized rate constants were then used to model the systems with 1 X loW3and 3 X M tripolyphosphate. Model simulations are compared with experimental observations in Figure 9. The model not only fits the system with 3 X M tripolyphosphate quite well (Figure 9B) but also successfullypredicts an increase followed by a slow decrease in soluble iron concentration in the system with 1X M tripolyphosphate. For even lower and higher M, tripolyphosphate concentrations, 8 X lo4 and 4 X this model still gave reasonably good predictions (Figure 9C,D). It is not surprising that the quantitative predictions were not always excellent, since factors such as changes in the reactive site density (which might decrease during the reaction period), electrostatic interactions, and lateral interactions between surface molecules were not considered. Indeed, it has been established that the adsorption of polyphosphates can be greatly affected by surface negative charge (21). The model would be able to simulate the kinetic results for systems containing total tripolyphosphate greater than 1X M reasonably well even without considering the adsorption of complexes (kll and k12in Figure 7). Similar kinetic curves have been reported in the literature fairly often. However, if kll and k12are removed from the model, it is impossible to reproduce the data or even the pattern in systems with total tripolyphosphate less than or equal to 1 X M, i.e., under conditions where there is a period with a net negative dissolution rate. For instance, when the data from the system containing 1 X M tripolyphosphate are used to determine the best-fit rate constants (Figure lo), the model simulates the experimental results very well when kll and k I z are included. However, it cannot simulate the observed trend of an increase followed by a slow decrease in soluble iron concentration when kll and kI2 are excluded. As noted earlier, the reason for this can be understood by inspecting reactions 5 (k5)and 6 (ks) in Figure 7. The model construct is such that as the

I I!

n9

I

0

+

0

+

0

Simulation

0 0

0.4

o,3]

* * +

1 2 $

,

-

pTzz-1 +

0

Obwvalion Simulation

$11

I MI

:in

Zoo

Time lminl I1 7

on

I

I

I

I

I

0

in

IO11

Im

?Do

Time 0 "

I6

Figure 9. Simulation results for systems with lo4, 3 X lo3, 8 X io4, and 4 X lo4 M tripolyphosphate. The rate constants for the simulation were optimized based on the system with 2 X M P3010(Figure 8.)

Optmiration-1 Optimization-2

I

1

I1

Fr- - k Llprcc"rsorc"mplel8 /

1

1 I1

,

1

?

,

1

-

/

1 ?!

,

/

,

1

,

I?

!(I

i

1

f,

/

1

16

I4

1

,

IX

,

1

1

20

Time imini

,

_-___

I

- 1

'I

I

*IC1

I

I

I

I\

I 1

I

I 1

'

I \

'

I

'

I

'

I YI

I

1s:

'

I 110

'

I 1-5

'

1

?M

T me lmin!

Figure 11. Model-predicted sorbed (A and 6) and dissolved (C) tripolyphosphate speciation for a system containing M P,Ot0. (B) is the same as (A), with an expanded scale from 0 to 20 min.

-

10 min, and the concentration of Fe-L reaches negligible values by t = -40 min. As each precursor complex leaves the surface, dissolved L and dissolved FeL compete to sorb to the newly available site. The sorption of LFe (in the form Fe-1-LFe) temporarily stops the dissolution at that site, while sorption of L initiates a new dissolution sequence. Soon, most surface sites are occupied by FeL (as Fe-1-LFe). The dissolution reaction slows because free ligands in solution are unable to compete effectively with FeL for these sites. The whole process can be treated as the conversion of Fe-I-L plus L in the early stages of reaction ( t < 10 min) to Fe-1-LFe plus FeL in the later stages ( t > 25 min); in the end, all the adsorbed tripolyphosphate is in the form Fe-1-LFe, and all the soluble tripolyphosphate is present as FeL. The previous discussion described systems where total tripolyphosphate was less than that necessary to saturate the surface. Figure 12 presents typical model output for 132

Environ Sci. Technol., Vol. 24, No. 1, 1990

",
I"

n

Figure 12. Model-predicted sorbed (A and B) and dissolved (C) tripolyphosphate speciation for a system containing 2 X lo3 M P,O,,. (6)is the same as (A), with an expanded scale from 0 to 20 min.

the alternative case, where total tripolyphosphate is greater than that necessary to saturate the surface. The early stages of the reaction are not very different in the two cases. Fe-I-L is the predominant surface species up to -10 min (Figure 12B). However, the concentration of Fe-I-L peaks and then starts to decrease after 2 min, while Fe-1-LFe increases (Figure 12B). As the surface is gradually enriched in Fe-1-LFe a t the expense of Fe-I-I,, the release of iron is retarded. After 1 h, almost all surface sites available for binding ligands are occupied by complexes in the form Fe-1-LFe (Figure 12A). From this point on, dissolution is very slow, proceeding only to the extent that free L in solution can occupy a free site during the short interval between desorption of LFe from a surface site and resorption of the same or a different LFe complex from solution. This process proceeds continuously until, at equilibrium, all the ligands in the system, both in solution (Figure 2) and at the surface, are complexed. The dissolution of ferrihydrite in the presence of tripolyphosphate is a typical example of the situation where

-

0

2i